Reversible Logic-MUX-Multiplier based Face Recognition using Hybrid Features.

Abstract

The face recognition is used to identify a person for access into work place, home, electronic gadgets etc. In this paper, we propose Reversible Logic MUX-Multiplier (RLMM) based Face Recognition using Hybrid Features. The novel concept of RLMM is introduced in designing multiplier for edge detection of face images using reversible logic gates. The steganography is used to convert two images of same person into one image. By using Canny Edge Detection, the edge detected stegoimages are obtained in order to create the database. The resize and Gaussian filter techniques are used in preprocessing of face images. The Discrete Wavelet Transform (DWT) and Local Binary Pattern (LBP) coefficients are computed from images and are fused to extract initial features. The final features are obtained from the output of Self Organizing Map (SOM) for the input of initial features. The Euclidian Distance (ED) is used to compare final features of face database images and test images to compute performance parameters. It is observed that the performance of the proposed method is better compared to existing methods. The proposed Multiplier and canny edge detection is coded using VHDL, synthesized and simulated using Xilinx ISE 14.5i tool. The face recognition is performed using Matlab 2012